From: Chris Wilson Date: Fri, 3 Feb 2017 12:57:17 +0000 (+0000) Subject: drm/i915: Remove overzealous fence warn on runtime suspend X-Git-Url: https://git.stricted.de/?a=commitdiff_plain;h=e0ec3ec698851a6c97a12d696407b3ff77700c23;p=GitHub%2FLineageOS%2Fandroid_kernel_motorola_exynos9610.git drm/i915: Remove overzealous fence warn on runtime suspend The goal of the WARN was to catch when we are still actively using the fence as we go into the runtime suspend. However, the reg->pin_count is too coarse as it does not distinguish between exclusive ownership of the fence register from activity. I've not improved on the WARN, nor have we captured this WARN in an exact igt, but it is showing up regularly in the wild: [ 1915.935332] WARNING: CPU: 1 PID: 10861 at drivers/gpu/drm/i915/i915_gem.c:2022 i915_gem_runtime_suspend+0x116/0x130 [i915] [ 1915.935383] WARN_ON(reg->pin_count)[ 1915.935399] Modules linked in: snd_hda_intel i915 drm_kms_helper vgem netconsole scsi_transport_iscsi fuse vfat fat x86_pkg_temp_thermal coretemp intel_cstate intel_uncore snd_hda_codec_hdmi snd_hda_codec_generic snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_timer snd mei_me mei serio_raw intel_rapl_perf intel_pch_thermal soundcore wmi acpi_pad i2c_algo_bit syscopyarea sysfillrect sysimgblt fb_sys_fops drm r8169 mii video [last unloaded: drm_kms_helper] [ 1915.935785] CPU: 1 PID: 10861 Comm: kworker/1:0 Tainted: G U W 4.9.0-rc5+ #170 [ 1915.935799] Hardware name: LENOVO 80MX/Lenovo E31-80, BIOS DCCN34WW(V2.03) 12/01/2015 [ 1915.935822] Workqueue: pm pm_runtime_work [ 1915.935845] ffffc900044fbbf0 ffffffffac3220bc ffffc900044fbc40 0000000000000000 [ 1915.935890] ffffc900044fbc30 ffffffffac059bcb 000007e6044fbc60 ffff8801626e3198 [ 1915.935937] ffff8801626e0000 0000000000000002 ffffffffc05e5d4e 0000000000000000 [ 1915.935985] Call Trace: [ 1915.936013] [] dump_stack+0x4f/0x73 [ 1915.936038] [] __warn+0xcb/0xf0 [ 1915.936060] [] warn_slowpath_fmt+0x5f/0x80 [ 1915.936158] [] i915_gem_runtime_suspend+0x116/0x130 [i915] [ 1915.936251] [] intel_runtime_suspend+0x64/0x280 [i915] [ 1915.936277] [] ? dequeue_entity+0x241/0xbc0 [ 1915.936298] [] pci_pm_runtime_suspend+0x55/0x180 [ 1915.936317] [] ? pci_pm_runtime_resume+0xa0/0xa0 [ 1915.936339] [] __rpm_callback+0x32/0x70 [ 1915.936356] [] rpm_callback+0x24/0x80 [ 1915.936375] [] ? pci_pm_runtime_resume+0xa0/0xa0 [ 1915.936392] [] rpm_suspend+0x12d/0x680 [ 1915.936415] [] ? _raw_spin_unlock_irq+0x17/0x30 [ 1915.936435] [] ? finish_task_switch+0x88/0x220 [ 1915.936455] [] pm_runtime_work+0x6f/0xb0 [ 1915.936477] [] process_one_work+0x1f3/0x4d0 [ 1915.936501] [] worker_thread+0x48/0x4e0 [ 1915.936523] [] ? process_one_work+0x4d0/0x4d0 [ 1915.936542] [] ? process_one_work+0x4d0/0x4d0 [ 1915.936559] [] kthread+0xd9/0xf0 [ 1915.936580] [] ? kthread_park+0x60/0x60 [ 1915.936600] [] ret_from_fork+0x22/0x30 In the case the register is pinned, it should be present and we will need to invalidate them to be restored upon resume as we cannot expect the owner of the pin to call get_fence prior to use after resume. Fixes: 7c108fd8feac ("drm/i915: Move fence cancellation to runtime suspend") Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98804 Reported-by: Lionel Landwerlin Signed-off-by: Chris Wilson Cc: Daniel Vetter Cc: Imre Deak Cc: Jani Nikula Cc: # v4.10-rc1+ Link: http://patchwork.freedesktop.org/patch/msgid/20170203125717.8431-1-chris@chris-wilson.co.uk Reviewed-by: Joonas Lahtinen --- diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 276676b2bbc8..78191e00a00c 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -2475,6 +2475,7 @@ static int intel_runtime_resume(struct device *kdev) * we can do is to hope that things will still work (and disable RPM). */ i915_gem_init_swizzling(dev_priv); + i915_gem_restore_fences(dev_priv); intel_runtime_pm_enable_interrupts(dev_priv); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index f3b5cb497921..495f1176d45e 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -2011,8 +2011,16 @@ void i915_gem_runtime_suspend(struct drm_i915_private *dev_priv) for (i = 0; i < dev_priv->num_fence_regs; i++) { struct drm_i915_fence_reg *reg = &dev_priv->fence_regs[i]; - if (WARN_ON(reg->pin_count)) - continue; + /* Ideally we want to assert that the fence register is not + * live at this point (i.e. that no piece of code will be + * trying to write through fence + GTT, as that both violates + * our tracking of activity and associated locking/barriers, + * but also is illegal given that the hw is powered down). + * + * Previously we used reg->pin_count as a "liveness" indicator. + * That is not sufficient, and we need a more fine-grained + * tool if we want to have a sanity check here. + */ if (!reg->vma) continue;